A frequency deviation estimation method for satellite VDES system

Abstract

A frequency offset estimation method for a satellite VDES system, characterized in that it includes the following steps: step 1, down-converting a carrier-containing signal output by an AD to move the signal to a baseband; step 2, performing CIC down-sampling processing on the down-converted signal to reduce the data rate to 4 times the symbol rate; step 3: performing frame synchronization processing on the CIC down-sampled signal to find the position of the frame header; step 4: performing coarse frequency offset estimation and compensation on the signal after the frame synchronization processing; step 5: performing Mpart fine frequency offset estimation on the signal after the coarse frequency offset compensation.

Description

Technical Field

[0001] The present invention relates to the field of satellite communication technology, and in particular to a frequency deviation estimation method for a satellite VDES system. Background Art

[0002] VDES (VHF Data Exchange System) is a system for strengthening and upgrading the ship automatic identification system (AIS) in the field of marine mobile services. In the VDES system, VDE is the core of the system. VDE includes a satellite communication system (VDE-SAT system), which enables remote two-way data exchange between satellites and ships.

[0003] In the satellite communication system (VDE-SAT system), the relative motion between the satellite and the ship will cause the existence of Doppler frequency shift in the signal. If there is no sufficiently accurate frequency deviation estimation technology to compensate for the frequency deviation, the satellite's receiving performance will be affected.

[0004] Chinese patent application CN107483078A proposes a method for estimating the receiving frequency deviation of the ASM system of the ship VDES system. The method performs phase detection on the signal after low-pass filtering to obtain the phase detection error, and caches the phase detection error in RAM, performs variable rate processing to adapt to FFT fast calculation; performs maximum peak detection on the output data of FFT calculation, finds the subscript of the maximum value, and calculates the frequency deviation estimation value. In this invention, the number of calculation points of FFT has a certain influence on the result. Once the number of calculation points is improperly selected, the error of frequency deviation estimation will be relatively large.

[0005] The paper "Research on Frequency Offset Estimation and Phase Tracking Algorithm in VDES" (Fan Jiahui et al., Electronic Measurement Technology, 2020) proposed a two-step M&M frequency offset estimation and phase tracking algorithm, which first makes a coarse frequency offset estimation with a larger range, then makes a fine frequency offset estimation with higher accuracy, and finally maintains the phase tracking state. This algorithm uses the autocorrelation function of the pilot sequence in the received signal to

[0006] The phase difference result of the number is smoothed and the smoothed result is used as the frequency estimation value. Although this algorithm can improve the estimation accuracy under low signal-to-noise ratio and improve the error performance of the demodulation system,

[0007] But it is only suitable for links with LINK ID.

[0008] The paper "Research on Frequency Offset Estimation Algorithm of ASM Signal in VDES" (Zhu Hongming et al., Electronic Measurement Technology, 2020) proposes a two-step estimation algorithm based on the M&M algorithm, which first makes a coarse frequency offset estimation with a larger range, and then makes a fine frequency offset estimation with higher accuracy. That is, the data-assisted M&M algorithm is first used to remove the large frequency offset, and then the M&M blind estimation algorithm is used to make a high-precision estimate to remove the residual small frequency offset. Although the fine frequency offset estimation method in this paper has a high frequency offset accuracy, its maximum estimated range is reduced to 1 / 4 of the original.

[0009] In order to solve the above problems, the present invention proposes a frequency offset estimation method for a satellite VDES system, which improves the frequency offset estimation accuracy by combining differential correlation coarse frequency offset estimation with Mpart fine frequency offset estimation, thereby improving the receiving performance of the satellite VDES system. Summary of the invention

[0010] The object of the present invention is to provide a method for estimating frequency deviation of a satellite VDES system, comprising the steps of:

[0011] Step 1: down-convert the signal containing the carrier output by the AD and move the signal to the baseband;

[0012] Step 2, performing CIC downsampling processing on the down-converted signal to reduce the data rate to 4 times the symbol rate;

[0013] Step 3: Perform frame synchronization processing on the CIC downsampled signal to find the position of the frame header;

[0014] Step 4: Perform coarse frequency offset estimation and compensation on the signal after frame synchronization processing;

[0015] Step 5: Perform Mpart fine frequency offset estimation on the signal after coarse frequency offset compensation.

[0016] Preferably, in step 1, the received AD sampled signal is multiplied with two local carriers, one in quadrature and one in phase, respectively, and after passing through a low-pass filter, the double frequency is filtered out to obtain two channels of orthogonal down-converted frequency data.

[0017] Preferably, in step 3, the frame synchronization processing includes differential detection calculation, data correlation calculation and threshold calculation.

[0018] Preferably, the differential detection calculation includes:

[0019] Since the data rate is four times the symbol rate after CIC downsampling, delaying one symbol time Ts requires delaying the sampled data by four sampling point times, that is, Ts=4*Tds.

[0020] in, Sampling frequency f ds =ps *4;

[0021] The downsampled signal is:

[0022]

[0023] Where b(n) is the baseband signal oversampled by 4 times, f d is the Doppler frequency deviation, T ds is the sampling period, For phase.

[0024] After delaying 4 sampling points, the conjugate of the delayed signal and the received signal are multiplied to obtain the differential result.

[0025] The downsampled signal is differentially detected, and the differential signal obtained is:

[0026]

[0027] Where z(n) is the baseband signal after 4 times oversampling and difference, f d To multiply the frequency deviation, T ds is the sampling period.

[0028] Preferably, the data correlation calculation includes: after differential detection, using the training sequence to generate a four-times oversampled local differential training sequence copy zs(n) and the differential data rd(n) to perform conjugate correlation:

[0029]

[0030] Where, l is the length of the local training sequence copy, and N is the length of the data after difference;

[0031] Since the training sequence has excellent autocorrelation characteristics after differentiation, when the received signal matches the training sequence, the correlator outputs a peak value, and the location of the correlation peak is the timing position;

[0032] The peak value of the correlator is:

[0033]

[0034] Among them, f d To multiply the frequency deviation, T ds is the sampling period.

[0035] Preferably, in actual application, the threshold calculation has a large signal dynamic range, and a constant false alarm detection method is used to obtain the threshold value.

[0036] Preferably, in step 4,

[0037] Coarse frequency offset estimation: The coarse frequency offset estimation module is used to preliminarily estimate the frequency offset in the received signal; the angle of the maximum correlation peak of the differential correlation is calculated to obtain the phase of the correlation peak, and the frequency offset is calculated based on the phase value.

[0038] Find the angle of the maximum correlation peak max{C(k)} and get the phase pcorr of the correlation peak:

[0039] p corr = arg·max{C(k)} = 2πf d *4T ds

[0040] Among them, f d To multiply the frequency deviation, T ds is the sampling period;

[0041] The frequency deviation f calculated based on the phase value d :

[0042]

[0043] Among them, p s is the symbol rate.

[0044] Preferably, in step 4,

[0045] Frequency offset compensation: After receiving the rough frequency offset estimation result, DDS is used to generate sine and cosine waves with the rough frequency offset estimation value to compensate the received signal for frequency offset. DDS needs to input a phase control word with a bit number of 32 bits. The frequency offset estimation value is calculated from the received phase value:

[0046]

[0047] Among them, phase is the received phase value, R S is the symbol rate;

[0048] Then the phase control word is calculated by the formula:

[0049] M=f*2 32 / f sys

[0050] Among them, f sys is the system clock, and f is the estimated frequency offset.

[0051] Preferably, in step 5,

[0052] Mpart fine frequency offset estimation takes the available data training sequence as an example. After the training sequence is correlated, a coarse estimated frequency offset is obtained, and this frequency offset in the original received data is compensated. Then, based on the position of the correlation peak of the training sequence, that is, the timing position, the training sequence is taken out, and then the training sequence is divided into M segments of data, and the Mpart correlation is performed using the data of these M segments of symbols.

[0053] Preferably, in step 5,

[0054] When M is 3: The specific steps related to 3part are:

[0055] Step 51: Take 3 available data segments: Take 3 data segments m1, m2 and m3 in the training sequence, and the corresponding received data are aS1(n), aS2(n), aL1(n);

[0056] Step 52: Calculate the correlation peak value of each segment of data: calculate the correlation peak value M1 of m1, the correlation peak value M2 of m2, and the correlation peak value M3 of m3;

[0057]

[0058] Where ≥f is the residual frequency offset, i.e., the fine frequency offset estimate, T S is the symbol period;

[0059] Step 53: Calculate the phase difference: calculate the first phase difference p0 by multiplying the conjugate of M1 and M2, and calculate the second phase difference p1 by multiplying the conjugate of M2 and M3;

[0060]

[0061] Where ≥f is the residual frequency offset, i.e., the fine frequency offset estimate, T S is the symbol period;

[0062] Step 54: Calculate the mean of the phase difference: calculate the mean of p0 and p1;

[0063] Step 55: Calculate the phase according to the mean value;

[0064]

[0065] Step 56: Calculate the frequency deviation according to the phase value.

[0066]

[0067] Among them, f s is the symbol rate.

[0068] The technical advantage of the present invention is that after using the coarse frequency offset estimation related to the differential training sequence, the Mpart related method is used to further perform fine frequency offset estimation on the data after the coarse frequency offset compensation. By combining the differential related coarse frequency offset estimation and the Mpart fine frequency offset estimation method, the frequency offset estimation accuracy is improved, thereby improving the receiving performance of the satellite VDES system. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 Frequency offset estimation flow chart;

[0070] Figure 2 Differential detection block diagram;

[0071] Figure 3 Related processing block diagram;

[0072] Figure 4 Mpart fine frequency offset estimation flow chart. DETAILED DESCRIPTION

[0073] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0074] The present invention proposes a frequency deviation estimation method for a satellite VDES system, and the specific steps are as follows:

[0075] Step 1: Down-convert the signal containing the carrier output by the AD and move the signal to the baseband;

[0076] Step 2: Perform CIC downsampling on the down-converted signal to reduce the data rate to 4 times the symbol rate;

[0077] Step 3: Perform frame synchronization processing on the CIC downsampled signal to find the position of the frame header;

[0078] Step 4: Perform coarse frequency offset estimation and compensation on the frame synchronized signal;

[0079] Step 5: Perform Mpart fine frequency offset estimation on the signal after coarse frequency offset compensation.

[0080] The above-mentioned frequency deviation estimation method of a satellite VDES system, wherein in the step 1, the received AD sampled signal is multiplied with the orthogonal and in-phase two-way local carrier respectively, and after passing through a low-pass filter, the double frequency is filtered out to obtain two-way orthogonal down-converted frequency data (no separate low-pass filter is designed here, because the subsequent data will pass through a CIC filter, which can filter out the redundant double frequency while downsampling).

[0081] In the step 2, a five-stage cascaded CIC filter is used to implement downsampling, and the data rate after downsampling is 4 times the symbol rate.

[0082] In step 3, the frame synchronization module is completed by differential detection, data correlation and threshold calculation. The down-sampled data is subjected to differential detection to obtain a differential result. The data after differential detection is subjected to data correlation. When the received signal matches the training sequence, the correlator outputs a peak value, so a threshold value can be set. When the correlator output peak value is higher than the threshold value, it can be considered that the frame header is detected, that is, the frame is synchronized.

[0083] In step 4, the correlation peak of the differential training sequence is used to perform coarse frequency offset estimation, and the coarse frequency offset compensation receives the coarse frequency offset estimation result and uses DDS to generate two orthogonal waveforms to perform frequency offset compensation on the data.

[0084] In step 5, the training sequence data is read, and the read data is used to perform Mpart fine frequency offset estimation.

[0085] The present invention will be further described below in conjunction with the accompanying drawings.

[0086] Step 1: Down-convert the signal containing the carrier output by the AD and move the signal to the baseband;

[0087] The down-conversion module receives the AD sampled signal and multiplies it with the orthogonal and in-phase local carriers respectively. After passing through the low-pass filter, the double frequency is filtered out to obtain two channels of orthogonal down-conversion data. (There is no separate low-pass filter designed here, because the data will pass through the CIC filter later, which can filter out the redundant double frequency while down-sampling)

[0088] Step 2: Perform CIC downsampling on the down-converted signal to reduce the data rate to 4 times the symbol rate;

[0089] Downsampling is to extract data. Since the data needs to be downsampled to 4 times the symbol rate, for example, 480 times is required. It is difficult to design a filter to extract such a high multiple at one time, so a five-stage cascaded CIC filter is used to divide 480 times into 2*3*4*4*5.

[0090] Step 3: Perform frame synchronization processing on the CIC downsampled signal to find the position of the frame header;

[0091] The frame synchronization module is composed of three parts: differential detection, data correlation and threshold calculation.

[0092] Differential detection: See the attached differential detection block diagram Figure 2 As shown in Figure 1, since the data rate is 4 times the symbol rate after CIC downsampling, delaying one symbol time Ts requires delaying the sampled data by 4 sampling point times T ds , that is, Ts = 4*T ds .

[0093] in, Sampling frequency f ds =p s *4.

[0094] The downsampled signal is:

[0095]

[0096] Where b(n) is the baseband signal oversampled by 4 times, f d is the Doppler frequency deviation, T ds is the sampling period, For phase.

[0097] After delaying 4 sampling points, the conjugate of the delayed signal and the received signal are multiplied to obtain the differential result.

[0098] The downsampled signal is differentially detected, and the differential signal obtained is:

[0099]

[0100] Where z(n) is the baseband signal after 4 times oversampling and difference, f d To multiply the frequency deviation, T ds is the sampling period.

[0101] Data correlation: After differential detection, the four-times oversampled local differential training sequence copy zs(n) generated by the training sequence is conjugate correlated with the differential data rd(n):

[0102]

[0103] Wherein, l is the length of the local training sequence copy, and N is the length of the data after difference.

[0104] Since the training sequence has excellent autocorrelation characteristics after differentiation, when the received signal matches the training sequence, the correlator outputs a peak value, and the position of the correlation peak is the timing position.

[0105] The peak value of the correlator is:

[0106]

[0107] Among them, f d To multiply the frequency deviation, T ds is the sampling period.

[0108] See the attached data related module processing block diagram Figure 3 As shown in the figure, since the correlation operation and the convolution operation only differ by a sequence fold, convolution can be used to achieve correlation. And since the filter essentially implements the convolution operation, the filter can be used to achieve the correlation operation.

[0109] Threshold calculation: In practical applications, the signal dynamic range is very large, and the constant false alarm detection method is used to obtain the threshold value.

[0110] Step 4: Perform coarse frequency offset estimation and compensation on the signal after frame synchronization processing;

[0111] Coarse frequency offset estimation: The coarse frequency offset estimation module is used to preliminarily estimate the frequency offset in the received signal. The angle of the maximum correlation peak of the differential correlation is calculated to obtain the phase of the correlation peak. The frequency offset can be calculated based on the phase value.

[0112] Find the angle of the maximum correlation peak max{C(k)} and get the phase pcorr of the correlation peak:

[0113] p corr = arg·max{C(k)} = 2πf d *4T ds

[0114] Among them, f d To multiply the frequency deviation, T ds is the sampling period.

[0115] The frequency deviation f calculated based on the phase value d :

[0116]

[0117] Among them, p s is the symbol rate.

[0118] Frequency offset compensation: After receiving the rough frequency offset estimation result, DDS is used to generate sine and cosine waves with the rough frequency offset estimation value to compensate the received signal for frequency offset. DDS needs to input a phase control word with a 32-bit phase control word. The frequency offset estimation value is calculated from the received phase value:

[0119]

[0120] Among them, phase is the received phase value, R S is the symbol rate.

[0121] Then the phase control word is calculated by the formula:

[0122] M=f*2 32 / f sys

[0123] Among them, f sys is the system clock, and f is the estimated frequency offset.

[0124] Step 5: Perform Mpart fine frequency offset estimation on the signal after coarse frequency offset compensation.

[0125] After using the differential correlation to estimate the coarse frequency offset, the Mpart correlation method is used to further perform fine frequency offset estimation on the data after the coarse frequency offset compensation.

[0126] The flowchart of Mpart fine frequency offset estimation is attached. Figure 4 , take the training sequence with 48*4 symbols as an example, after the training sequence is correlated, a rough estimate of the frequency offset can be obtained, and the frequency offset in the original received data is compensated, and then the 48*4 symbols are taken out according to the position of the correlation peak of the training sequence, that is, the timing position, and then the training sequence is divided into three segments of 16*4 data, and these three segments of 16*4 symbols are used to perform 3part correlation (it can also be 4part or 5part correlation, you can calculate it reasonably by yourself, the principle is the same).

[0127] The specific steps related to 3part are:

[0128] (1) Take three available data segments: Take three data segments m1, m2 and m3 in the training sequence, and the corresponding received data are aS1(n), aS2(n), aL1(n), and the total length is N = 192;

[0129] (2) Calculate the correlation peak value of each segment of data: calculate the correlation peak value M1 of m1, the correlation peak value M2 of m2, and the correlation peak value M3 of m3;

[0130]

[0131] Where ≥f is the residual frequency offset, i.e., the fine frequency offset estimate, T S is the symbol period.

[0132] (3) Calculating the phase difference: The first phase difference p0 is calculated by multiplying the conjugate of M1 and M2, and the second phase difference p1 is calculated by multiplying the conjugate of M2 and M3;

[0133]

[0134] Where ≥f is the residual frequency offset, i.e., the fine frequency offset estimate, T S is the symbol period.

[0135] (4) Calculate the mean of the phase difference: find the mean of p0 and p1;

[0136] (5) Find the phase based on the mean;

[0137]

[0138] (6) Calculate the frequency deviation based on the phase value.

[0139]

[0140] Among them, f s is the symbol rate.

[0141] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as a limitation of the present invention. After reading the above content, it will be apparent to those skilled in the art that various modifications and substitutions of the present invention will occur. Therefore, the protection scope of the present invention should be limited by the appended claims.

Claims

1. A frequency deviation estimation method for a satellite VDES system, characterized in that: Includes steps: Step 1: down-convert the signal containing the carrier output by the AD and move the signal to the baseband; Step 2, performing CIC downsampling processing on the down-converted signal to reduce the data rate to 4 times the symbol rate; Step 3: Perform frame synchronization processing on the CIC downsampled signal to find the position of the frame header; Step 4: Perform coarse frequency offset estimation and compensation on the signal after frame synchronization processing; Step 5: Perform Mpart fine frequency offset estimation on the signal after coarse frequency offset compensation.

2. The frequency deviation estimation method of a satellite VDES system according to claim 1, characterized in that: In step 1, the received AD sampled signal is multiplied with the orthogonal and in-phase local carriers respectively, and after passing through a low-pass filter, the double frequency is filtered out to obtain two channels of orthogonal down-converted frequency data.

3. The frequency deviation estimation method of a satellite VDES system according to claim 1, characterized in that: In step 3, the frame synchronization processing includes differential detection calculation, data correlation calculation and threshold calculation.

4. The frequency deviation estimation method of a satellite VDES system according to claim 3, characterized in that: The differential detection calculation includes: Since the data rate is four times the symbol rate after CIC downsampling, delaying one symbol time Ts requires delaying the sampled data by four sampling point times, that is, Ts=4*Tds. in, Sampling frequency f ds =p s *4; The downsampled signal is: Where b(n) is the baseband signal oversampled by 4 times, f d is the Doppler frequency deviation, T ds is the sampling period, For phase. After delaying 4 sampling points, the conjugate of the delayed signal and the received signal are multiplied to obtain the differential result. The downsampled signal is differentially detected, and the differential signal obtained is: Where z(n) is the baseband signal after 4 times oversampling and difference, f d To multiply the frequency deviation, T ds is the sampling period.

5. The frequency deviation estimation method of the satellite VDES system according to claim 3, characterized in that: Data related calculations include: After differential detection, the four-times oversampled local differential training sequence copy zs(n) generated by the training sequence is conjugate correlated with the differential data rd(n): Where, l is the length of the local training sequence copy, and N is the length of the data after difference; Since the training sequence has excellent autocorrelation characteristics after differentiation, when the received signal matches the training sequence, the correlator outputs a peak value, and the location of the correlation peak is the timing position; The peak value of the correlator is: Among them, f d To multiply the frequency deviation, T ds is the sampling period.

6. The frequency deviation estimation method of a satellite VDES system according to claim 3, characterized in that: In practical applications, the signal dynamic range is very large, and the constant false alarm detection method is used to obtain the threshold value.

7. The frequency deviation estimation method of a satellite VDES system according to claim 1, characterized in that: In step 4, Coarse frequency offset estimation: The coarse frequency offset estimation module is used to preliminarily estimate the frequency offset in the received signal; the angle of the maximum correlation peak of the differential correlation is calculated to obtain the phase of the correlation peak, and the frequency offset is calculated based on the phase value. Find the angle of the maximum correlation peak max{C(k)} and get the phase pcorr of the correlation peak: p corr =arg·max{C(k)}=2πf d *4T ds Among them, f d To multiply the frequency deviation, T ds is the sampling period; The frequency deviation f calculated based on the phase value d : Among them, p s is the symbol rate.

8. The frequency deviation estimation method of a satellite VDES system according to claim 7, characterized in that: In step 4, Frequency offset compensation: After receiving the rough frequency offset estimation result, DDS is used to generate sine and cosine waves with the rough frequency offset estimation value to compensate the received signal for frequency offset. DDS needs to input a phase control word with a bit number of 32 bits. The frequency offset estimation value is calculated from the received phase value: Among them, phase is the received phase value, R S is the symbol rate; Then the phase control word is calculated by the formula: M=f*2 32 / f sys Among them, f sys is the system clock, and f is the estimated frequency offset.

9. The frequency deviation estimation method of a satellite VDES system according to claim 1, characterized in that: In step 5, Mpart fine frequency offset estimation takes the available data training sequence as an example. After the training sequence is correlated, a coarse estimated frequency offset is obtained, and this frequency offset in the original received data is compensated. Then, based on the position of the correlation peak of the training sequence, that is, the timing position, the training sequence is taken out, and then the training sequence is divided into M segments of data, and the Mpart correlation is performed using the data of these M segments of symbols.

10. The frequency deviation estimation method of a satellite VDES system according to claim 9, characterized in that: In step 5, When M is 3: The specific steps related to 3part are: Step 51: Take 3 available data segments: Take 3 data segments m1, m2 and m3 in the training sequence, and the corresponding received data are aS1(n), aS2(n), aL1(n); Step 52: Calculate the correlation peak value of each segment of data: calculate the correlation peak value M1 of m1, the correlation peak value M2 of m2, and the correlation peak value M3 of m3; Where ≥f is the residual frequency offset, i.e., the fine frequency offset estimate, T S is the symbol period; Step 53: Calculate the phase difference: calculate the first phase difference p0 by multiplying the conjugate of M1 and M2, and calculate the second phase difference p1 by multiplying the conjugate of M2 and M3; Where ≥f is the residual frequency offset, i.e., the fine frequency offset estimate, T S is the symbol period; Step 54: Calculate the mean of the phase difference: calculate the mean of p0 and p1; Step 55: Calculate the phase according to the mean value; Step 56: Calculate the frequency deviation according to the phase value. Among them, f s is the symbol rate.

Images